Number Of Input Parameters Research Articles

DICTIONARY-BASED DETERMINISTIC METHOD OF GENERATION OF TEXT CORPORA

This paper examines the problem of solving software engineering tasks in developing information systems for natural language processing. Generating corpora of text data is highlighted as a specific task of this problem. An analysis of the basic CorDeGen method was carried out, which is one of the corpus generation methods specially developed for this specific problem. This study shows that this method has a limited scope due to the use of “artificial” terms to fill the texts.The paper proposes a new modified DBCorDeGen method that solves this shortcoming thanks to the use of an additional dictionary of terms that is supplied to the input of the method. The DBCorDeGen method preserves most of the characteristic features of the basic method, which are important for its use in solving software engineering tasks: determinism, speed of operation (including the possibility of combining with parallel modification), the possibility of a priori description of the structure and properties of the generated corpus. The only disadvantage compared to the basic method is the increase in the number of input parameters, however, compared to other methods of generating corpora presented in the literature, it is relatively small, and due to it, the scope of application of corpora generated by this method significantly increases.As an experimental test of the proposed modified DBCorDeGen method, the task of sentiment analysis of the texts of the generated corpus is considered. The study shows that when using the basic CorDeGen method, it is impossible to obtain sentiment analysis results different from neutral polarity for all texts. When using the proposed method, it is possible to obtain different results using different dictionaries. Thus, it is confirmed that the proposed DBCorDeGen method has a larger scope than the basic method.

Artificial intelligence-driven enhanced CBR modeling of sandy soils considering broad grain size variability

The soil packing, influenced by variations in grain size and the gradation pattern within the soil matrix, plays a crucial role in constituting the mechanical properties of sandy soils. However, previous modeling approaches have overlooked incorporating the full range of representative parameters to accurately predict the soaked California bearing ratio (CBRs) of sandy soils by precisely articulating soil packing in the modeling framework. This study presents an innovative artificial intelligence (AI)-based approach for modeling the CBRs of sandy soils, considering grain size variability meticulously. By synthesizing extensive data from multiple sources, i.e. extensive tailored testing program undertaking multiple tests and extant literature, various modeling techniques including genetic expression programming (GEP), multi-expression programming (MEP), support vector machine (SVM), and multi-linear regression (MLR) are utilized to develop models. The research explores two modeling strategies, namely simplified and composite, with the former incorporating only sieve analysis test parameters, while the latter includes compaction test parameters alongside sieve analysis data. The models' performance is assessed using statistical key performance indicators (KPIs). Results indicate that genetic AI-based algorithms, particularly GEP, outperform SVM and conventional regression techniques, effectively capturing complex relationships between input parameters and CBRs. Additionally, the study reveals insights into model performance concerning the number of input parameters, with GEP consistently outperforming other models. External validation and Taylor diagram analysis demonstrate the GEP models' superiority over existing literature models on an independent dataset from the literature. Parametric and sensitivity analyses highlight the intricate relationships between grain sizes and CBRs, further emphasizing GEP's efficacy in modeling such complexities. This study contributes to enhancing CBRs modeling accuracy for sandy soils, crucial for pertinent infrastructure design and construction rapidly and cost-effectively.

#2534 Classification of patients with acute kidney damage on the base of blood pressure value as a risk factor using a neural network

Abstract Background and Aims In acute kidney damage, the monitoring of hemodynamic status is best achieved by determining the blood pressure value. It is considered that the values of diastolic blood pressure in young and healthy people are constant from the aorta to the periphery, and the values of systolic and pulse pressure increase from the periphery to the aorta. The frequency of hypertension correlates with the presence of a cardiovascular precipitating risk factor in acute kidney injury, hypertension is also a modifying risk factor in the second and third stages of acute kidney injury, while the presence of hypotension on admission can be a clinical indicator of its cause. In patients with acute kidney damage and in patients with sepsis, the mechanism of autoregulation of mean arterial pressure, which is affected by stroke volume and heart rate, is not effective. The aim of the paper is to examine the classification of hospitalized patients with acute kidney injury based on blood pressure as a risk factor. Method The study included 86 patients over 18 years of age of both sexes (39 men 45.3% and 47 women 54.7%) with an average age of 66.85 ± 15.30 years who were diagnosed with acute kidney injury on admission to hospital treatment. Patients were divided into groups according to the stages of renal insufficiency. The study was designed as a cross-sectional comparative study. The assessment of the existence of acute kidney damage, as well as the determination of the stage of the disease, was based on the diagnostic criteria proposed by the K/DIGO expert group for acute kidney damage. The study did not include patients who had disease progression after acute kidney failure and were treated with dialysis for more than three months. Basic clinical, biochemical and hemodynamic parameters were determined in all subjects at the beginning of treatment, except for the analysis of nitrogen status, which was determined both before and at the end of treatment. During the formation of the neural network (pattern recognition Mat Lab), the parameters of systolic arterial pressure, diastolic arterial pressure normalized by min-max normalization to values from 0 to 1 were included. The data were randomly distributed into three categories: 80% for training, 10% for test, 10% for validation, which means that the parameters of 68 patients were used for training, 9 patients for validation and another 9 patients for the test. The number of input parameters is 33 to 39, while the number of output parameters is 2 in relation to treatment. Formed neural networks with different included blood pressure parameters, structures of 25 neurons in the hidden layer, with 55 and 15 epochs and with accuracy of 97.7% and 90.7% of patient classification. Results Our results show that 40.3% of patients in the third stage of acute kidney injury have hypertension. The lowest average values of systolic (108.44 ± 22.81 mmHg) and diastolic (65.56 ± 15.29 mmHg) blood pressure were experienced by patients in the first stage of acute kidney damage, and the lowest values of mean arterial pressure existed in the second stage of acute kidney damage (68.20 ± 44.96 mmHg) .Statistically significant risk factors for the development of the third stage of acute kidney damage in the multivariate analysis were patients older than 65 years, and mean arterial pressure values less than 65 mmHg. The confusion matrices showed that the classifiers adapted to the minimum and maximum values of systolic, diastolic and mean arterial blood pressure are well adapted and achieve satisfactory accuracy with 97.7% and 90.7%. Conclusion By comparing blood pressure parameters in hospitalized patients in relation to the stage of acute kidney damage, no statistically significant deviations were found. Univariate logistic regression analysis showed that age over 65 was a significant independent risk factor for stage three acute kidney injury, while the multivariate model showed that risk factors for the development of stage three acute kidney injury were patients same age and middle arterial blood pressure less than 65 mmHg. Satisfactory accuracy of formed neural networks when classifying patients during validation training and test was also demonstrated.

Reduced-order modelling for real-time physics-based variation simulation enhanced with adaptive sampling and optimized interpolation

AbstractManufacturing industries require a right-first-time paradigm to remain competitive. Variation simulation (VS) is a key tool to predict variation of the final shape of flexible assemblies, allowing to reduce defects and waste. VS models involving compliant sheet-metal parts commonly integrate physics-based simulation with statistical approaches (usually Monte Carlo simulation). Although increasingly used as a backbone of synthesis techniques for (stochastic) optimization of assembly systems, the main roadblock of VS methods is the intense computational costs due to time-intensive simulations and high-dimensional design space. Driven by the need of time reduction, this paper presents an innovative real-time physics-based VS model of assembly systems with compliant sheet-metal parts. The proposed methodology involves a non-intrusive reduced-order model (niROM), empowered by a novel adaptive sampling procedure for dataset generation, and a cross-validation-based optimized radial basis function (RBF) formulation for interpolation. Demonstrated through two case studies—(i) a remote laser welding operation to predict mechanical distortions, with two input parameters, and (ii) the assembly of an aircraft vertical stabilizer with five input parameters—the methodology achieves accurate real-time results, with up to a 43% improvement in accuracy compared to traditional sampling techniques. Findings highlight the critical influence of the sampling strategy and the number of input parameters on ROM accuracy. Better results are reached by employing adaptive sampling in combination with optimum RBF, which additionally disengages the user from the choice of the interpolation settings. This study unlocks new avenues in the field of variation simulation and dimensional/quality monitoring by narrowing the gap between any advanced CAE solver and VS models with real-time physics-based simulations.

Machine learning optimization for catalytic desulfurization of petroleum: Multi-layered perceptron, Multi Task Lasso, and Gaussian process regression models

Machine learning models were developed and optimized in this study to optimize the process of hydrodesulfurization (HDS) for fuel production. A number of input parameters were considered and optimized to find the best response parameters. The input parameters of process/model included: pressure, catalyst dosage, temperature, and feed’s sulfur content. This paper focuses on optimizing and evaluating AdaBoost ensemble models for predicting sulfur concentration, emission percentage, and HDS cost in industrial processes for fuel treatment. Employing Multilayer Perceptron (MLP), Gaussian Process Regression (GPR), and Multi-Task Lasso (MTL) as base models, and utilizing Tabu Search for tuning of hyper-parameter, we report the performance of the best models. The Bagged GPR stands out with a significant R2 of 0.967 for sulfur concentration and 0.959 for HDS cost, accompanied by RMSE values of 1.7691E+02 and 1.4191, respectively. For emission prediction, Bagged MTL demonstrates notable performance with an R2 of 0.877 and RMSE of 8.4235. These outcomes highlight the efficacy of ensemble models in enhancing predictive accuracy and underscore the potential for informed decision-making in industrial settings for the area of petroleum industry.

Modeling and simulation of liquid metal battery (LMB) materials development using statistical package for social sciences (S.P.S.S.)

There are a number of input parameters that are considered in relation to the stimulatory possibility of constructing a Liquid Metal Battery (LMB). This paper talks about the modeling approach possible for use in LMB research work. Equivalent Circuit Modeling (E.C.M.) is the most common method used to analyze input data or parameters. In analyzing some of the basic elements, such as electrical capacitance, electrical resistance, open circuit voltage, terminal voltage, temperature, time response, time constants, State of Charge (SOC), etc. The cell impedance could be calculated by predicting the system elements that would play key roles in the determination of the parameter identification method for the battery’s Equivalent Circuit Model. Secondly, each element in the model has a known behaviour, which mainly depends on the element type and the values of the parameters that characterize that element. In EIS software, the Graphical Model Editor could be used to build an equivalent circuit model, or the befitting model could be carefully and properly selected. Thirdly, in fitting the Equivalent Circuit Model (ECM) to the initial data or parameters, one must take note that the parameters are strongly dependent on temperature, heat, and losses. Such are: the Open Circuit Voltage (OCV), which is strongly dependent on the temperature, the loss processes depend on temperature; losses produced by the loss processes are dissipated as heat energy; the heat generated or consumed by the electrochemical reactions during normal operation has a defined temperature; and a system with a defined temperature window with safe, stable, and efficient operation is achievable. At the end of this research, the simulation of Lithium (Li) and Cadmium (Cd) was found to be in the proportion of 67:33, which is used to determine the strength of the reactivity of the metals. It can be informed in this article that Bi and Li chemical compositions of the metals are equal to 49 for Li and 51 for Bismuth, which makes the overall reactivity very high, which could be used for LMB development.

Analytical models for flow time estimation of additive manufacturing machines

The use of Additive Manufacturing (AM) technology has largely increased in the last years. Because of its large differences from conventional technologies, the use of AM in production systems might call for new strategies in production planning and control. To this aim, this paper proposes analytical models to predict aggregate performance measures such as flow time, work in process, and production throughput, for production systems characterised by Laser Powder Bed Fusion AM technology. These indicators could be used both in operations strategy development and in technology comparison. The proposed models differentiate for their detail of the analysis and the number of input parameters that need to be estimated. The results show that the level of detail of the model affects the analysis leading to quite different values of the performance measures, especially in the case of highly saturated systems. Also, a discussion about the applicability of the proposed model to other AM technologies show whether and to what extent the proposed models can be applied for modelling other AM technologies.

Computational approach for structural and thermal behavior of laser-machined micro-grooves on alumina ceramic using ANSYS

Heat affected zone on the micromachined material, a hindrance in achieving good precision and accuracy, which are adequate requirements for micromachined materials, is a drawback of nanosecond micromachining assisted by laser. The primary objective of this paper is to take into consideration a number of input parameters, such as the pulse width and the temperature of the laser beam, and study the effect that these parameters have on the heat flux generated as well as on the temperature distribution along the square grooves of 200 µm width and depth that are to be micro-machined on an alumina ceramic workpiece. However, processing of alumina ceramic on a micron-scale requires careful consideration to raise the quality standard. This will be accomplished by analyzing the data collected from the micro-machining process. Using the ANSYS® software, it is possible to visualize not only the region where heat is generated but also the way in which the input parameters influence the structural qualities of the micro machined groove surface. The many thermal models that have been constructed have revealed specific changes in the structural qualities that have had a significant impact on the upper width, the lower width, and the depth of the groove when applied to a variety of different parametric settings. Finally, developed analytical model is compared with the experimental models. The highest prediction error was found to be 1.68%, while, for constant 50 W power and 150–250 mm/s of speed, the average prediction error was determined to be 1.18%. As a result, the proposed model closely matches the experimental findings, and the proposed model can be used to estimate channel profile.

Physics-Informed Machine Learning for Surrogate Modeling of Heat Transfer Phenomena

Abstract In this paper, we propose a sparse modeling method for automatically creating a surrogate model for nonlinear time-variant systems from a very small number of time series data with nonconstant time steps. We developed three machine learning methods, namely, (1) a data preprocessing method for considering the correlation between errors, (2) a sequential thresholded non-negative least-squares method based on term size criteria, and (3) a solution space search method involving similarity model classification—to apply sparse identification of nonlinear dynamical systems, as first proposed in 2016, to temperature prediction simulations. The proposed method has the potential for wide application to fields where the concept of equivalent circuits can be applied. The effectiveness of the proposed method was verified using time series data obtained by thermofluid analysis of a power module. Two types of cooling systems were verified: forced air cooling and natural air cooling. The model created from the thermofluid analysis results with fewer than the number of input parameters, predicted multiple test data, including extrapolation, with a mean error of less than 1 K. Because the proposed method can be applied using a very small number of data, has a high extrapolation accuracy, and is easy to interpret, it is expected not only that design parameter can be fine-tuned and actual loads can be taken into account, but also that condition-based maintenance can be realized through real-time simulation.

Evaluating the most significant input parameters for forecasting global solar radiation of different sequences based on Informer

The number of existing global solar radiation (GSR) observation stations is limited, and it is challenging to meet the demand for scientific research and production. Different forecasting horizons of solar radiation correspond to various applications. Therefore, it is critical to design realistic models to predict the GSR of varying sequence lengths. This study proposes a prediction model based on the analysis of the Pearson correlation between GSR and each input parameter and the establishment of different input combinations related to the result of Pearson analysis in four high-quality datasets. The proposed Informer model compares the results with five classical machine learning models on four datasets with R2, RMSE, and skill score (S) as evaluation metrics. This study examines the proposed model's prediction performance for five prediction lengths, four climate zones, three sampling frequencies, and two input types. The results showed that the Informer model performs well with the clearness index and pressure as the input. Besides, the RRMSE values are less than 10% under optimal input in long sequence forecasting. The findings suggested that the proposed advanced Informer model is a reliable alternative for GSR prediction due to its high predictive accuracy under diverse prediction lengths, sampling frequencies, climate zones, and the number of input parameters.

Ultimate bearing capacity prediction method and sensitivity analysis of PBL

The ultimate bearing capacity of Perfobond leiste (PBL) is one of the key parameters to evaluate the bearing capacity and reliability of steel–concrete structures, and it is very important to predict the ultimate bearing capacity of PBL more accurately. Based on the improved cuckoo search algorithm (CS), two prediction model optimized by back propagation neural network (BPNN) algorithm and extreme learning machine (ELM) were proposed. The local search ability of CS algorithm was improved by the triangular mutation operator and the distance-based distributed discovery probability, and the global search ability was improved by multi-step selection strategy. The weight, threshold, number of input parameters and number of nodes in the hidden layer of BPNN and ELM were optimized by the triangular multi-step cuckoo search (TMCS). The comprehensive sensitivity analysis (CSA) method and Morris sensitivity analysis (MSA) method were used to analyze the sensitivity of six key parameters of PBL, such as thickness of perforated steel plate, diameter of perforated holes, number of perforated holes, diameter of through reinforcement, yield strength of through reinforcement and compressive strength of concrete. The experimental data of push-out tests in published literatures were selected as samples, and the results show that the proposed TMCS-ELM algorithm and TMCS-BPNN algorithm can accurately predict the ultimate bearing capacity of PBL, and the average errors are 4.17% and 2.16% respectively. The sensitivity analysis results show that the compressive strength of concrete has the highest influence on the bearing capacity of PBL, followed by the yield strength of through reinforcement.

Outlier-Robust Surrogate Modeling of Ion-Solid Interaction Simulations.

Data for complex plasma-wall interactions require long-running and expensive computer simulations. Furthermore, the number of input parameters is large, which results in low coverage of the (physical) parameter space. Unpredictable occasions of outliers create a need to conduct the exploration of this multi-dimensional space using robust analysis tools. We restate the Gaussian process (GP) method as a Bayesian adaptive exploration method for establishing surrogate surfaces in the variables of interest. On this basis, we expand the analysis by the Student-t process (TP) method in order to improve the robustness of the result with respect to outliers. The most obvious difference between both methods shows up in the marginal likelihood for the hyperparameters of the covariance function, where the TP method features a broader marginal probability distribution in the presence of outliers. Eventually, we provide first investigations, with a mixture likelihood of two Gaussians within a Gaussian process ansatz for describing either outlier or non-outlier behavior. The parameters of the two Gaussians are set such that the mixture likelihood resembles the shape of a Student-t likelihood.

Deep learning model for temperature prediction: A case study in New Delhi

AbstractThis study is based on temperature prediction in the capital of India (New Delhi). We have adopted different ML models such as (MPR and DNN) which are designed and implemented for temperature prediction. The MPR models are varied on the degree of the polynomial, whereas the DNN models differ in the number of input parameters. DNNM‐1 takes date, time, and temperature as input, and DNNM‐2 receives date, time, temperature, pressure, humidity, and dew point as input parameters, whereas DNNM‐3, is a complex model that takes date, time, temperature, pressure, humidity, dew point, and 32 weather conditions as input. To evaluate the accuracy of the predictions, a comparison of the predicted temperature and the actual recorded temperature is done, and the performance and accuracy of the models are examined. The MPR models work well in case of fewer input features, but as the number of input features grows, the DNN model outperforms the MPR models. The DNN model (DNNM‐3) outperformed the other models with better accuracy as compared to past evidence.

Global sensitivity study of non-reactive and sorptive solute dispersivity in multiscale heterogeneous sediments

Lagrangian-based transport models provide effective ways of understanding mass transport processes within aquifer systems. The models provide a direct relationship between sparse data on sedimentary architecture (e.g., facies proportions and mean lengths) and physical and geochemical sediment properties (e.g., hydraulic conductivity (K) and sorption distribution coefficient (Kd)) to transport observables such as dispersion. Data sparsity leads to parameter uncertainty, which in turn makes model prediction uncertain. This study identifies the key uncertain inputs for both non-reactive and sorptive solute dispersivity through a global sensitivity analysis. Estimates of the individual and correlation contributions of input parameters to model output are provided. Data from two sites with different scales of heterogeneities are used to evaluate the sensitive parameters of non-reactive dispersivity at different scales. The results show that sorptive solute dispersivity is most sensitive to in-facies mean Kd, followed by Kd variance, while non-reactive plume dispersivity is most sensitive to in-facies mean K, followed by the volume proportions and mean lengths of facies types. When the heterogeneity integral scale increases to 102-103m, hydraulic gradient becomes a non-negligible factor controlling the non-reactive solutes transport. The convergence of the sensitivity indices and the effect of different sampling methods on the results are also evaluated in this study. The results show that the number of input parameters and the complexity of the model determine the sampling size to achieve the ranking convergence of the sensitive indices. Sobol sequence sampling scheme outperforms in terms of convergence rate and accuracy over the Monte Carlo and Latin Hypercube sampling schemes. The results of this study will improve our understanding of the complex model system, and also provide guidance for further field investigation and data collection.

MIA-NDN: Microservice-Centric Interest Aggregation in Named Data Networking.

The named data networking (NDN)-based microservice-centric in-network computation poses various challenges in terms of interest aggregation and pending interest table (PIT) lifetime management. A same-named microservice-centric interest packet may have a different number of input parameters with nonidentical input values. In addition, the same-named interest packet with the same number of parameters may have different corresponding parameter values. The vanilla NDN request aggregation (based on the interest name, while ignoring the input parameters count and/or their corresponding values) may result in false aggregation. Moreover, the microservice-centric requested computations may fail to accomplish in the default 4s PIT timer due to the input size. To address these challenges, this paper presents MIA-NDN: microservice-centric interest aggregation in named data networking. We designed microservice-centric interest-naming to enable name-based communication. MIA-NDN develops a robust interest aggregation mechanism that not only performs the interest aggregation based on the interest name but also considers the input parameter counts and their corresponding values in the interest aggregation process to avoid false packet aggregations. A dynamic PIT timer mechanism based on input size was devised that avoids the PIT entry losses if the execution time exceeds the default PIT timer value to avoid computation losses and uphold the application quality of service (QoS). Extensive software-based simulations confirm that the MIA-NDN outperforms the benchmark scheme in terms of microservice-centric interest aggregation, microservice satisfaction rate, and communication overhead.

Derivatives of feed-forward neural networks and their application in real-time market risk management

Market risk management of financial derivatives requires the efficient calculation of their price sensitivities with respect to changes in market factors. This paper shows how a deep feed-forward neural network which has been trained for pricing derivative instruments can be efficiently used to calculate these sensitivities as well. The proposed method is a fast and easily implementable alternative approach to automatic differentiation, and it simultaneously calculates all the first- and second-order derivatives of a multilayer feed-forward neural network with respect to its input features. The paper quantifies the performance improvement of the proposed method over a recent, publicly available implementation of automatic differentiation for a wide range of network sizes. The number of input parameters in these networks corresponds to those of commonly used financial models with stochastic volatility. The numerical accuracy of the proposed sensitivity calculations is demonstrated with a case study, calculating price sensitivities of European options under stochastic volatility. While the paper focuses on financial applications, the results presented herein are applicable to all deep feed-forward neural networks with sufficiently smooth activation functions.

Evaluation of Ground-Based Models for Estimating Surface Albedo with In-Situ Radiometric Measurements across China

Surface albedo is an essential parameter in many solar radiation applications. Although several models are available, it remains debatable whether they are applicable to other locations. Using long-term daily measurements of radiation acquired by observational networks in China, this study examined the applicability of six existing albedo models: Ineichen model (IeM), Gueymard model (GM), Dong model (DeM), Iziomon-Mayer model (IMM), Morton model (MM), and Zhou model (ZeM). The evaluation results of model performance through statistical analysis showed that among the available ground albedo models, ZeM had the best overall performance at 12 selected stations, IeM was shown to provide acceptable estimations for locations where albedo records are readily available. The statistical results of individual station have shownthat the number of input parameters is not the only key factor for determining the predictive performance of ground albedo models. In other words, a simple model has potential for accurate estimation of ground albedo with appropriate model parameters. Therefore, the simple two-parameter DeM was selected to re-calibrate with in-situ radiometric measurements, which can be adopted as a surrogate for ZeM to predict surface albedo in China.

Optimization and off-design calculations of a turbojet engine using the hybrid ant colony – particle swarm optimization method

PurposeThe purpose of the paper is to present component matching and off-design calculations using generic components maps.Design/methodology/approachMulti objective hybrid optimization code is integrated with turbojet function code. Both codes are developed for the research study. Initially, methodology is applied on a numerical propulsion system simulation (NPSS) example engine cycle calculations. Effect of matching constants are shown. Later, component matching and application is done on JetCat engine. Calculations are compared with measured test data. And additional operating conditions are calculated using the matched component constants.FindingsObtained matching constants provided very good results with NPSS example and also JetCat test measurements. Optimization algorithm is practical for turbojet engine component matching and off-design calculations. Off-design matching provides information about the turbine and exhaust areas of an unknown turbine engine. Thus it is possible to perform off design calculations at various operating conditions. Finding detailed turbine maps is difficult than finding compressor maps. In that case characteristic turbine curve may be a good alternative.Research limitations/implicationsSelected component maps and the target engine components should be similar characteristics. For a one/two stage turbine, characteristic curves can be applied. Validation should be extended on different type of compressor and turbines.Practical implicationsOperators and researchers usually need more information about the available turbojet engines for increasing the effective usage. Generally, manufacturers do not provide such detailed information to public. This study introduces an alternative methodology for engine modeling by using generic component maps and thus obtaining information for off-design calculations. User is flexible for selecting/scaling the compressor and turbine maps.Originality/valueA hybrid optimization code is used as a new approach. It can be used with other engine functions; for instance functions corresponding to turboshaft or turbofan engines, by modifying the engine function. Number of input parameters and objective functions can be modified accordingly.

Estimated ultimate recovery prediction of fractured horizontal wells in tight oil reservoirs based on deep neural networks

Accurate estimated ultimate recovery prediction of fractured horizontal wells in tight reservoirs is crucial to economic evaluation and oil field development plan formulation. Advances in artificial intelligence and big data have provided a new tool for rapid production prediction of unconventional reservoirs. In this study, the estimated ultimate recovery prediction model based on deep neural networks was established using the data of 58 horizontal wells in Mahu tight oil reservoirs. First, the estimated ultimate recovery of oil wells was calculated based on the stretched exponential production decline model and a five-region flow model. Then, the calculated estimated ultimate recovery, geological attributes, engineering parameters, and production data of each well were used to build a machine learning database. Before the model training, the number of input parameters was reduced from 14 to 9 by feature selection. The prediction accuracy of the model was improved by data normalization, the early stopping technique, and 10-fold cross validation. The optimal activation function, hidden layers, number of neurons in each layer, and learning rate of the deep neural network model were obtained through hyperparameter optimization. The average determination coefficient on the testing set was 0.73. The results indicate that compared with the traditional estimated ultimate recovery prediction methods, the established deep neural network model has the strengths of a simple procedure and low time consumption, and the deep neural network model can be easily updated to improve prediction accuracy when new well information is obtained. Cited as: Luo, S., Ding, C., Cheng, H., Zhang, B., Zhao, Y., Liu, L. Estimated ultimate recovery prediction of fractured horizontal wells in tight oil reservoirs based on deep neural networks. Advances in Geo-Energy Research, 2022, 6(2): 111-122. https://doi.org/10.46690/ager.2022.02.04

Dynamics of the Rolling Stock and the Choice of Parameters of Vibration Dampers

Modern railway rolling stock should meet requirements regarding comfort (maximum travel speed with minimum vibrations of wagons, noiselessness of movement, etc.).To eliminate the influence of dynamic loads, rolling stock is equipped with vibration dampers. The objective of the work is to select the parameters of the vibration dampers of rolling stock, depending on its characteristics, to ensure the due indicators of comfort and safety of transportation of passengers and goods by rail. To achieve this objective, applied methods of mathematical modelling were based on numerical programming of operation of dynamic systems. The indicators of vibration dampers are evaluated according to the results of studies of the dynamics of the rolling stock (in particular, of vibration protection rates).Assessment of dynamic state of the rolling stock implies application of methods of mathematical and physical modelling, which include the development of a physical and mathematical model, a calculation algorithm, and computer programming. The study of the mathematical model by numerical methods makes it possible to carry out a multifactorial experiment using a large number of input parameters (factors) and to select the characteristics of vibration dampers that are optimal for the conditions under consideration.To solve dynamic problems, the harmonic perturbation model, which is the most widespread, was specified in the form of a sinusoid with a period corresponding to the rail length.A quantitative assessment of the vibration process (frequency, amplitude) makes it possible to identify the main processes occurring in the system under consideration under various types of external load. The introduced assumptions related to rigidity, mobility and geometric immutability of the system allow determining the methods for obtaining a mathematical model and considering the vibrations as flat ones.The equations were solved in MathCad Prime 4.0 package using the Runge–Kutta method with automatic step selection. The subsequent study of the properties of the dynamic system was carried out by changing the resistance parameter of dampers of the first stage of spring suspension, while recording the values of the amplitude of the vibrations of the system and the period.The analysis of the results has shown that the vibration period of the body and bogies under any changes in the resistance parameter of the damper remains unchanged, while rational parameters of resistance of axle box dampers have been revealed for specified indicators. Hydraulic vibration dampers with the revealed parameters used on rolling stock help to reduce wear and damageability of running gears, improve ride comfort and traffic safety, as well as to reduce repair and maintenance costs.